You wrote:
"I know I won't get a degree out of this and I doubt I can get published"
Sorry to disagree! :-)!
For getting published, try some journal on human-computer interaction. Since you know much more about your field than I do, look for other journal 'varieties' as well.
For what you are doing, that is, a lot of independent, call it, R&D, if you get results anything like what you want, then it would be surprising if you didn't have something publishable. If this argument is not enough, then look at much of what does get published and conclude that much of it is not very high quality stuff!
Uh, one way to improve your paper a notch or two over what is common is just to write well, say, well organized, clear, including writing good English with good spelling, punctuation, and grammar.
Once you are published at least once, better, say, three times, sorry, but you should be able to get a Ph.D.! :-)! Maybe this is a big disappointment, but it's true!
How then to get a Ph.D? Four points:
(1) At at least some of the best US research universities, there is no coursework requirement for a Ph.D.
(2) The requirement for a dissertation is, say, "an original contribution to knowledge worthy of publication". Since you've already been there, done that, got the T-shirt, there's no question. Big, huge advantage.
(3) One more requirement would likely be the qualifying exams. These exams are to show that you are 'qualified' to move on to research and do research. But, uh, did I mention that you've already been there, done that, got the T-shirt? The difficulty of the qualifying exams varies widely, but generally the 'polish on your halo' can be important in deciding who passes.
So, it can be good to have some high quality halo polish; here are three: (A) Some of the best halo polish is published papers, the more papers, more highly regarded, the better the polish. (B) More halo polish is when it looks like you might be a successful entrepreneur who gets good publicity for the university and, maybe later on, is, uh, 'thankful and generous'! (C) More good polish is that you actually did this work, conceived it, took it on, got it done, independently, which means you are promising as a good researcher for the future bringing more good publicity to the university.
(4) There may be a requirement for 'residency' for a year. This can mean that maybe you show up on campus some day in September and again in the spring to defend your dissertation which you submit by taking a stack of your published papers and putting a big staple in the UL corner.
For more:
You should pick the university and department carefully. You might try, say, the bio-engineering program at Johns Hopkins. I would focus on such higher end engineering programs with 'biology' contact. Of course try MIT and Cal Tech. Don't settle for Southern Sawgrass U.
You may be able to get a research grant, before, during, or after your Ph.D. program: At your university, ask the people who know about grant sources. Maybe the US DoD VA would give you a grant. A research grant is the 'magic bullet' to rapid progress in academics because you are bringing money to the university! You will understand better when you see what fraction of the money the university keeps for 'overhead'! Did I hear someone say, "Money talks."?
Uh, one way to pick a department and prof: Pick your journal carefully! That is, if you want to get into program A with full prof P editor of relevant journal X, then submit your paper to journal X and, if it is accepted, then, uh, take 'advantage' of this contact you made with prof P to get your Ph.D. Note: It can be possible for prof P to 'direct' your dissertation at a school not his.
Uh, once your paper is accepted, prof P may invite you to present your paper at his conference on Computer-Human Interaction or some such. Likely accept! Then that's two bullets on your CV.
Note: When you submit your paper, likely you do not have to give any significant biographical information at all. So, you don't need a high school transcript! In particular, it doesn't have to be clear if you have a Ph.D. or not. So, before your Ph.D., you are fully 'qualified' to submit a paper.
Note: Commonly a journal plays pocket pool with your paper for a year before sending you results of reviews. So, two rejections and one acceptance would take three years. Bummer.
So, to speed up the process, go online, find maybe 50 appropriate journals, to the editor in chief of each, via e-mail or on paper, write a nice, one page letter outlining your paper and asking "might your journal be interested?", enclose a copy of your paper, and, then, only from the responses make a 'formal submission'.
Uh, the research universities make a big, huge deal out of 'research' both for the faculty and for at least the Ph.D. students. Else they'd have to know something about the real world and teach it, right? Horrors!
Uh, the research universities long since concluded that by a wide margin the most difficult part of a Ph.D. and the usual point of failure is just the research for the dissertation. For some students, that work is hard enough to threaten their life (literally). For others, it's put their feet up for an afternoon, think up some good stuff, write it up, send it in, get it accepted for publication, and shout "Done!".
In my opinion, you've already passed the main obstacle: You've decided to do some independent work.
For more, you've picked a problem, have some productive lines of attack, and are making progress.
For more, if you get the research results you have in mind, then you will have something "new, correct, and significant": "New"? No doubt what you are working on does not yet exist. "Correct"? It works! "Significant"? Ask Hawking or many people and/or groups working with the handicapped. E.g., ask the US DoD VA.
You can do something cute here: You satisfy "significant" because you solve a practical problem and not because you have a theorem or counterexample that settles some old conjecture about differential cohomology that only six people in the world know about. Readers listen up: This example makes a useful, general point!
Presto: "New, correct, and significant" are the usual criteria for publication!
That is, you've already given up on taking a problem from a prof, pleasing a prof, caring mostly about what will please a prof, looking for praise, approval, status, prestige, guidance, 'mentoring' from a prof, etc.
But, but, but, your work would never get venture funding, right? Wrong! One of the 'themes' Brad Feld likes to pursue is human-computer interaction, especially without a traditional keyboard! Uh, if you had Hawking, the DoD VA, and some organization for the handicapped on your side, had some beta testers, ..., then you might get a venture capital check -- it's not hopeless.
Sorry to disagree with your:
"I know I won't get a degree out of this and I doubt I can get published". :-)!
There's an old quote: "Be wise; generalize.".
Okay: How I got a Ph.D.: I started with a practical problem I'd identified and worked on before grad school. I saw a solution intuitively. In my first year in grad school, I took some advanced math that let me turn my intuitive stuff into some solid theorems and proofs for a solution, and in my first summer independently I did that. That work was all but the software and typing for my dissertation.
For some interim 'halo polish', I saw a problem, thought for a few evenings, roughly saw a solution, and then signed up for a 'reading course' on that problem. I worked for a few more days and saw a much nicer solution, wrote it up, turned it in, and was done with the reading course in two weeks. The work looked publishable and was -- I published it later. So, in two weeks I'd created "an original contribution to knowledge worthy of publication", that is, satisfied the requirement for a dissertation. Good halo polish.
A student who does such independent work is "difficult", but on this point read the recent Fred Wilson blog at his AVC.COM.
You will not be the first to do some independent work and later get a Ph.D. for it. Uh, one of the biggest topics now in computer science is a 'good' algorithm as in the set of algorithms P as in the question P versus NP. Well, likely and apparently the formulator of a 'good' algorithm was Jack Edmonds. Can get a start on him at
http://en.wikipedia.org/wiki/Jack_Edmonds
Yes, he also won a von Neumann prize, etc. Little things like that!
As at Wikipedia, he was at University of Maryland (UM). Uh, as I heard the story, he was not a happy Ph.D. student at UM so left to the National Bureau of Standards (NBS) about 40 miles away. Then he published some good work in graph theory. Finally some of the math faculty at UM, feeling a little guilty, drove to NBS, smoked a peace pipe, and said essentially "Put a staple in the UL corner of your papers and we'll be pleased to call it a Ph.D. dissertation in our department.".
Final point: Of course, D. Knuth knows about academics. Well buried in 'The TeXBook' is:
"The traditional way is to put off all creative aspects until the last part of graduate school. For seventeen or more years, a student is taught 'examsmanship', then suddenly after passing enough exams in graduate school he's told to do something original."
Yes, there very much is a conflict here: The approach to getting into grad school can strongly conflict with the approach to writing a dissertation.